Having said that, the most likely form of a question that they can ask that requires a spreadsheet would be to ask for the p-value for a likelihood ratio test in which the test stat under the null is a 50-50 mixture of a chi^2 with d and d-1 degrees of freedom. (I.e., we are testing removing a random effect and are eliminating d parameters). In that case, if your test stat is x, you get the answer from:

= 0.5*CHISQ.DIST.RT(x, d) + 0.5*CHISQ.DIST.RT(x, d-1)

and the spreadsheet reminds you of the full name of the formula as well as the syntax as you type it.

]]>I’m thinking about adding some problems that nominally involve splines, but only in a superficial way. E.g., write down a spline model and ask what happens to the effective number of parameters if you reduce the SD on the priors. It will go down because that’s what happens with anything if you reduce the SD on the priors.

]]>Yes, there will be an updated formula sheet and flashcards.

]]>